Temperature-Controlled Direct Imprinting of Ag Ionic Ink: Flexible Metal Grid Transparent Conductors with Enhanced Electromechanical Durability

Next-generation transparent conductors (TCs) require excellent electromechanical durability under mechanical deformations as well as high electrical conductivity and transparency. Here we introduce a method for the fabrication of highly conductive, low-porosity, flexible metal grid TCs via temperature-controlled direct imprinting (TCDI) of Ag ionic ink. The TCDI technique based on two-step heating is capable of not only stably capturing the Ag ionic ink, but also reducing the porosity of thermally decomposed Ag nanoparticle structures by eliminating large amounts of organic complexes. The porosity reduction of metal grid TCs on a glass substrate leads to a significant decrease of the sheet resistance from 21.5 to 5.5 Ω sq−1 with an optical transmittance of 91% at λ = 550 nm. The low-porosity metal grid TCs are effectively embedded to uniform, thin and transparent polymer films with negligible resistance changes from the glass substrate having strong interfacial fracture energy (~8.2 J m−2). Finally, as the porosity decreases, the flexible metal grid TCs show a significantly enhanced electromechanical durability under bending stresses. Organic light‐emitting diodes based on the flexible metal grid TCs as anode electrodes are demonstrated.

at high-temperatures (>200 °C) to improve the electrical conductivity 23,24 . After high-temperature sintering process, the metal grid TCs are transferred from the glass substrate to the polymer substrate for the fabrication of flexible metal grid TCs.
Among the wet deposition methods, the DI of colloidal metal NPs can directly produce micro/nanoscale metal structures at low-costs and in a high-throughput manner without expensive etching steps and metal evaporation 25,26 . Recently, our group suggested a reservoir-assisted DI method based on an Ag ionic ink, which has several advantages such as higher metal content and less aggregation than colloidal metal NPs, for the fabrication of high-performance flexible metal grid TCs 17,18 . However, the Ag NP structures fabricated by the thermal decomposition of Ag ionic ink include numerous micro/nanoscale pores due to the elimination of organic complexes 27,28 . Unfortunately, those pores can deteriorate electrical and mechanical properties of the flexible metal NP structures [29][30][31][32][33] . First, the formation of many pores precludes the generation of highly conductive metal grid TCs. Second, since high-temperature sintering process (~300 °C) induces strong interfacial fracture energy (IFE) between the metal and the substrate, micro/nanoscale pores can cause serious damage to metal NP structures during the transfer process. Finally, as the cracks are initiated and propagated near numerous pores under mechanical deformations, the flexible metal grid TCs may be easily damaged by static or dynamic bending stresses. The formation of micro/nanoscale pores inside the metal NP structures should be minimized in the development of flexible optoelectronic applications.
In order to resolve this problem, we introduce a novel temperature-controlled direct imprinting (TCDI) process of Ag ionic ink based on two-step heating for the generation of highly conductive, low-porosity, flexible metal grid TCs. This TCDI technique led to stably capturing the Ag ionic ink and reducing the porosity of thermally decomposed Ag NP structures. The electrical resistivity (ρ m ) and R s of the metal grid TCs at a fixed transmittance (T 550nm ) at 550 nm were represented as a function of the organic complex contents and geometrical calculation. The effect of porosity on the transfer process of metal grid TCs to the polymer film was explored at different sintering temperatures. In addition, the effect of porosity on an electromechanical durability of the flexible metal grid TCs was examined under the static and dynamic bending stresses. Finally, the utility of flexible metal grid TCs as anode electrodes was demonstrated by fabricating of OLEDs.

Results
Temperature-controlled direct imprinting of Ag ionic ink. The TCDI of Ag ionic ink, which involves the two-step heating, is schematically illustrated in Fig. 1a: (i) Ag ionic ink (10 μL) was imprinted on a fluorinated glass substrate using a grid-patterned mold under low pressure (P = 120 kPa) and low temperature; (ii) the first heating step was performed at an evaporation temperature (T E ) of 50 °C over 5 min to stably capture the Ag ionic ink inside the grid-patterned cavity. It should be noted that when the Ag ionic ink is initially heated at an increased T E , large amounts of the filled ink leave from the grid-patterned cavity through a liquid film between the mold and the substrate due to carbon dioxide generation during the thermal decomposition of Ag ions 34 ; (iii) for a liquid film with negligible thickness (h f ≈ 0), the second heating step was carried out at decomposition temperature (T D ) = 90 °C to eliminate organic complexes derived from the thermal decomposition of the Ag ions and to improve their thermal decomposition rate; (iv) after complete solvent evaporation, the grid-patterned mold was carefully removed from the substrate. The Ag NP structures were thermally treated at sintering temperature (T S ) = 300 °C for 10 min. Figure 1b shows a schematic illustration of (i) the two-step heating of TCDI and (ii) the one-step heating of typical DI, respectively. In Fig. 1b(iii-iv), focused ion beam-scanning electron microscope (FIB-SEM) images show a cross-section of the metal grid line structures fabricated using TCDI and DI of Ag ionic ink, respectively. As T D was increased from 50 °C to 90 °C, the micro/nanoscale pores in the metal grids were significantly reduced. This micro/nanoscale pores may be mainly generated by the elimination of organic complexes during the thermal decomposition process. Amounts of organic complexes inside the Ag NP structures were compared using thermogravimetric analysis (TGA) and derivative thermogravimetric (DTG) analysis of Ag NP-organic complex powders, respectively. Figure 1c shows the TGA and DTG curves obtained from Ag NP-organic complex powders after complete drying at T D = 50 and 90 °C, respectively. The Ag NP-organic complex powders dried at the lower T D showed the larger weight loss and the faster decomposition rate in the temperature range of 100-200 °C. The Ag NP-organic complex powders dried at T D = 50 and 90 °C showed a weight loss of 21.24 and 6.17%, respectively. These results indicated that the metal grids generated using the DI of Ag ionic ink may include a greater fraction of the organic complex by weight, about 15% more, than those obtained using the TCDI of Ag ionic ink. Figure 1d plots a weight loss of the Ag NP-organic complex powders dried at different values of T D . As T D increased from 50 to 100 °C, the organic complex contents of the powder decreased. The inset shows photographic images of the metal-organic complex powders. The Ag NP-complex powders dried at T D = 50 °C showed dark black due to high organic complex contents (left), while those obtained at T D = 90 °C showed bright silver due to relatively low organic complex contents (right).
Highly conductive, low-porosity metal grid TCs. Figure 2a shows a SEM image of the Ag NP structures fabricated over the grid-patterned mold area (10.5 mm × 10.5 mm). The presence of unwanted residual layers within the grid spacing was negligible. The inset shows a magnified SEM image of the Ag NP structures at the intersection of metal grids. A linewidth (~8 μm) of the metal grid was smaller than that of the original grid-patterned cavity (15 μm) due to mold deformation. Figure 2b shows the values of R s and T 550nm for the metal grid TCs generated using the TCDI of Ag ionic ink. As T D increased from 50 to 90 °C, R s of the metal grid TCs decreased significantly, holding T 550nm over 91.2%; however, the value of R s for the metal grid TCs fabricated at T D = 100 °C increased and fluctuated as solvent evaporation and thermal decomposition occurred near the boiling point of the base-solvent (toluene, 110.6 °C). Figure S1(a) shows transmittance spectra over a wavelength range of 350-800 nm of the metal grid TCs fabricated at different values of T D . The spectral transmittance of metal grid TCs fabricated at T D = 100 °C did significantly decrease from 350 to 500 nm. Also, figure of merit (FoM), which generally predicts the performances of TCs, was compared for the metal grid TCs generated at different values of T D in Figure S1(b). As T D increased to 90 °C, the FoM of the metal grid TCs was increased from 250 to 572. The value of ρ m of thermally reduced Ag NP line structures fabricated at different values of T D is compared in Fig. 2c. The value of ρ m could be expressed using the following relationship: where ρ o is the electrical resistivity of the thermally reduced metal structures without organic complex contents, w p is the weight percent of organic complex and k p is a correction factor from experimental data. The correction factor was determined to be k p = 0.0727. The value of ρ o (6.5287 × 10 −6 Ω cm) was 4 times larger than that of the bulk Ag (1.59 × 10 −6 Ω cm). The R s of the metal grid TCs was calculated based on the structural size and organic complex weight. The equation used to predict R s was constructed by modifying the expression for R s obtained using a square wire network, as suggested by Van de Groep et al. 13 . The cross-section of the Ag NP structures was regarded as trapezoidal, and R s was predicted using the following equation:  Transfer process of metal grid TCs. The metal grid TCs on the glass substrate should be transferred onto the polymer substrate for use in flexible optoelectronic devices. A transfer process of metal grid TCs provided the flexibility and significantly reduced the surface roughness that can cause optoelectronic device failure. Figure 3a shows a schematic illustration of the transfer process, based on a sandwich structure (glass substrate/polymer film/glass substrate) to generate a uniform, thin and transparent polymer film on both sides. It should be noted that when the polymer film becomes thinner, the flexible metal grid TCs show better electromechanical durability under the same bending radius (r) due to a decrease of the nominal bending strain 35 . The Norland Optical Adhesive 81 (NOA 81), which shows a high optical transparency over a wide spectral range with a good mechanical flexibility and a strong adhesion force to metal structures, is used as the polymer film. The NOA 81 solution was poured onto the metal grids-patterned glass substrate, and then was covered by the glass substrate. After the UV exposure, the metal grids embedded within NOA 81 film were detached from two glass substrates. The effect of the porosity reduction on the transfer process of metal grid TCs was explored by measuring the resistance changes of metal grid TCs sintered at different values of T S . The higher T S induces the more increase of the IFE between the metal and the glass substrate, which causes serious damage to the flexible metal grid TCs during the transfer process. Especially, the metal structures with many pores are prone to be broken due to the increase of the IFE. In Figure S2(a), the IFE of Ag NP film on the glass substrate was quantitatively measured using a double cantilever beam (DCB) fracture mechanics testing method at different values of T S . In Figure S2(b), the inset shows a schematic diagram of the DCB test specimen consisting of a fluorinated glass substrate (donor substrate), a Ag NP film, an adhesion layer, and the plasma-treated glass substrate (receiving substrate). Figure 3b plots the IFE between the Ag NP film and the substrate after T S over the range 200-300 °C. Although the glass substrate was treated by the fluorine, the IFE between the Ag NP film and the glass substrate depended strongly on T S . The Ag NP films sintered at T S = 300 °C, in particular, displayed 3-fold larger IFE (~8.2 J m −2 ) than those sintered at T S = 200 °C (~2.7 J m −2 ). Figure 3c shows the normalized resistance change (∆R/R o ) and R s of the flexible metal grid TCs after the transfer process of the metal grid TCs sintered T S = 200-300 °C. As T S increased, ∆R/R o of the flexible metal grid TCs based on the DI of Ag ionic ink (T D = 50 °C) significantly increased from  Figure 4a(ii) shows an atomic force microscopy (AFM) surface profile of the flexible metal grid TCs, which presented smooth surfaces with a root-mean-square surface roughness of 6.7 nm and a maximum peak-to-valley value of 43.8 nm. Figure 4b compares the values of T and R s for the flexible metal grid TCs with the values obtained from a commercially available ITO-coated polyethylene terephthalate (PET) film. Here, it should be noted that the transmittance spectra of the bare NOA 81 film, the flexible metal grid TCs, and the ITO-coated PET film include the transmittance through the substrate. The flexible metal grid TCs performed better (R s of 5.5 Ω sq −1 and T 550nm of 81.47%) than the ITO-coated PET film (R s of 15.0 Ω sq −1 and T 550nm of 78.35%). The T of the ITO-coated PET film fluctuated over the wavelength range of 350-800 nm, whereas the corresponding values of the flexible metal grid TCs remained constant.
The electromechanical durability of the flexible metal grid TCs under static and dynamic bending stresses is an important factor to be considered in flexible optoelectronic devices 35,36 . As the number of pores in the metal grid structures decreased, the mechanical durability of the flexible metal grid TCs can be improved by inhibiting pore-induced crack initiation and propagation under bending stresses [29][30][31][32][33] . This resulted in reducing the electrical resistance changes of the metal grid structures under bending stresses. Figure 4c plots the values of ∆R/R o obtained from the flexible metal grid TCs fabricated using TCDI and DI of Ag ionic ink under a static bending test. The value of ∆R/R o of the ITO-coated PET film under a static bending test was evaluated as a reference. As the flexible metal grid TCs were bent up to r = 1 mm, the values of ΔR/R o increased for these two flexible metal grid TCs (R o = 6.3 and 21.0 Ω sq −1 ) by a factor of 2.2 and 113.3 respectively. The R/R o of the flexible metal grid TCs fabricated using the DI of Ag ionic ink significantly increased at r = 1 mm due to serious damage. Figure 4d plots the values of ∆R/R o obtained from the flexible metal grid TCs fabricated using TCDI and DI of Ag ionic ink under a dynamic bending test with r = 7.5 mm. The value of ∆R/R o of the ITO-coated PET film under a dynamic bending test was evaluated as a reference. After the only several bending cycles, the resistance of the ITO-coated PET film significantly increased to a few tens of kΩ due to formation of numerous and visible cracks. After 1000 cycles of repeated bending/relaxation, ∆R/R o obtained from the flexible metal grid TCs (R o = 7.4 and 43.5 Ω sq −1 ) increased by a factor of 1.5 and 86.4, respectively. When T D increased, the porosity reduction of the flexible metal grids led to decreasing ∆R/R o under static and dynamic bending stresses (a significant decrease in the relative ratio of resistance change by more than 51 times at r = 1 mm and by more than 57 times after the 1000 bending cycles, respectively). The low-porosity, flexible metal grid TCs fabricated using the TCDI of Ag ionic ink showed

Discussion
Applications to flexible organic light-emitting diodes. Figure 5a shows a schematic diagram of OLEDs based on the flexible metal grid TCs. Indium zinc oxide (IZO) layers with a thickness of 50 nm were used to cover the flexible metal grid TCs, which exhibit R s of 5.5 Ω sq −1 and T 550nm of 81.47%. Poly(3,4ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) and MoO 3 layers were used on the top of IZO films for efficient hole injection. The 4,4′-Bis(N-carbazolyl)-1,1′-biphenyl (CBP) layer doped with green phosphorescent emitters of bis(2-(2-pyridinyl-N)phenyl-C)(acetylacetonate) iridium (III) (Ir(ppy) 2 acac, 7 wt%) was used as an emitting layer. The CBP and 2,2′,2″-(1,3,5-Benzinetriyl)-tris(1-phenyl-1-H-benzimidazole) (TPBi) layers were used as hole-and electron-transport layers, respectively. Figure 5b shows the normalized electro-luminescence spectrum of the flexible OLEDs in the normal direction in the wavelength range of 350-700 nm, which is similar to that of the ITO-based TCs. The inset shows the operating image of the flexible OLEDs under bending. The OLEDs fabricated on the flexible metal grid TCs show an electrically stable operation due to their relatively good uniformity and low roughness, as shown in Fig. 5c. The inset shows the angular emission profile, which was used to obtain the external quantum efficiency (η EQE ) and power efficiency (η PE ) of the devices. Therefore, the OLEDs based on the flexible metal grid TCs show the comparable values of η EQE and η PE with those of ITO-based OLEDs, as shown in Fig. 5d. The flexible OLEDs exhibit η EQE of 22.0% and η PE of 61.8 lm W −1 at 111.6 cd m −2 , while ITObased OLEDs show η EQE of 21.5% and η PE of 60.6 lm W −1 at 116.2 cd m −2 , respectively.
In summary, we demonstrated the fabrication of highly conductive, low-porosity, flexible metal grid TCs via the TCDI of Ag ionic ink. As T D increased during the thermal decomposition of Ag ionic ink, the porosity of metal grid structures was significantly reduced due to the elimination of large amounts of organic complexes. Also, both ρ m and R s of metal grid TCs were estimated based on the weight percent of organic complex and geometrical calculations. The porosity reduction led to improving the optoelectrical properties of metal grid TCs (R s < 6 Ω sq −1 at T 550nm = 91%) and preventing serious damage (∆R/R o < 6%) from the strong IFE between the metal and the substrate during the transfer process. In addition, we verified that the porosity reduction resulted in significantly enhancing an electromechanical durability of the flexible metal grid TCs under static and dynamic bending stresses. The OLED based on flexible metal grid TCs as anode electrodes was demonstrated. The uniformity, reliability and scalability of the metal grid TCs should be further improved by the combination of uniform pressing-stable detaching system and the exact control of fluidic properties (viscosity and surface tension) before this technology can be commercialized. We believe that this strategy can provide a useful approach to enhance optoelectrical properties and an electromechanical durability of solution-processed metallic TCs for the development of next-generation flexible optoelectronic devices.

Methods
Ag ionic ink. The Ag ionic ink (TEC-IJ-060, Inktec) consisted of Ag ions (Ag alkyl carbamate complexes), a base solvent (methanol and toluene) and additives. Prior to the TCDI of Ag ionic ink, methanol was evaporated at 100 °C for 10 min to improve the concentrated ink filling. The Ag alkyl carbamate complexes were decomposed to Ag NPs, carbon dioxide, and the corresponding alkyl amines by heating above 50 °C for a few minutes 34 . PDMS mold. The PDMS solution (Sylgard 184, Dow Corning) was generated by mixing the silicon elastomer kit and a curing agent (10:1), and this mixture was poured onto a SU-8 master. After PDMS curing at 100 °C for 1 hour, a PDMS mold was carefully released from the SU-8 master. In the PDMS mold, the grid-patterned cavity was designed with a width (w) of 15 μm and a spacing (s) of 250 μm by calculating the geometrical shadow zone, T = s 2 /(s + w) 2 . The value of a cavity height was limited to 7.5 μm to prevent the destruction of Ag NP structures during the detachment of PDMS mold. However, the value of T 550nm measured from the metal grid TCs was higher than that obtained using the geometrical calculation of grid-patterned mold due to mold deformation (93.2 and 89%, respectively).
Effect of pressure. The effect of P, ranged from 30 to 300 kPa, was examined by considering the Poiseuille's law (h f 4 is inversely proportional to ∆P) 37 . At 30 kPa, most of the ink was not captured inside the grid-patterned mold due to a large value of h f . The ink was effectively captured inside the mold cavity over 100 kPa. The pressure was optimized at 120 kPa by comparing Rs and T 550nm of the metal grid TCs. Also, the relatively uniform P based on horizontal levels led to decreasing both R s and resistance fluctuations of the metal grids.
Transfer process. All types of glass substrate, including soldalime glass and borosilicate glass, were treated using 1 H, 1 H, 2 H, 2H-perfluorooctyl-trichlorosilane (448931, Sigma-Aldrich) for 3 min in a vacuum chamber. After the sintering process, a metal grid-patterned glass substrate (soldalime) was re-treated with the fluorinated silane. The NOA 81 solution was poured onto the metal grid-patterned glass substrate, and it was covered by a borosilicate glass substrate. After UV exposure, the borosilicate glass substrate was detached from the NOA film, and then the metal grid-embedded within NOA 81 film (its thickness of 150 μm) was detached from the sodalime glass substrate.
Fabrication of flexible organic light-emitting diodes. The IZO film (~50 nm) was deposited on the flexible metal grid TCs using an RF sputtering system (RF power: 120 W). The PEDOT:PSS (Clevios PVP AI4083, Heraeus, 45 nm) layers were spin-coated on the top of the IZO films. PEDOT:PSS was coated also on pre-coated ITO glass substrates (<12 Ω sq −1 , AMG Inc., Korea) as a reference after O 2 plasma treatment (70 W, 1 min). The samples with PEDOT:PSS were dried to remove any residual solvent at 100 °C for 10 min on the hot plate. Finally, the PEDOT:PSS-coated samples were loaded into a thermal evaporator for deposition of organic, metal oxide layers, and metal electrodes under high vacuum (3 × 10 −6 Torr) conditions. The stacked multilayers of OLEDs were composed of MoO 3 (10 nm)/CBP (20 nm)/CBP doped with 7 wt% of Ir(ppy) 2 acac (20 nm)/TPBi (55 nm)/ LiF (1 nm)/Al (100 nm). DCB test. The Ag NP film/glass substrate (sodalime) specimens were fabricated with the size of 37.5 mm × 9 mm. The specimens were sandwiched by an additional glass beam using an epoxy (353ND, Epoxy Technology). The final structure of the DCB test specimen was the glass substrate/epoxy/Ag/glass substrate. The connection parts with the DCB test machine and aluminum loading tabs were attached to the specimens with the selective epoxy (DP420, 3M) and the specimens were cured at 120 °C for 1 h in a convection oven.